FRT 237: FY23 CARS Data Technical Assistance and Automation

Abstract FRT 237, FY23 CARS Data Technical Assistance and Automation From 2019 to 2020, Quality Counts drove the Kentucky state-maintained highway system using Rieker Inc’s CARS solution. While collecting CARS data, they also recorded video with a dash- mounted GoPro camera. The video was later used to extract a sign inventory. Quality Counts also processed the CARS data resulting in a signing plan for all detected curves. KTC enhanced this dataset by determining advanced warning signing needs at each curve according to MUTCD requirements and mapping the curves alongside Kentucky’s current warning sign inventory, allowing KYTC district offices to identify curves in need of signage updates. Additionally, AADT, crash data, and recent resurfacing dates were incorporated into the dataset. A web tool to filter the database was provided as well. However, all the above data and improvements will remain static unless a process is developed to update the database either in real time or annually. This research will provide updates to the enhanced CARS database using current year data as well as investigate methodologies to automate updates to the CARS database. Additionally, based on feedback from users, further updates and improvements may be applied to the database and web tool. Research Plan 1. Update the FY22 CARS data enhancements with current year (FY23) data (AADT, Crashes, Resurfacing dates, etc.) 2. Add summarized HERE speed data to database to show actual speeds driven through curves. 3. Survey KYTC district engineers for feedback on the CARS/warning sign map and PowerBI spreadsheet tool developed in the FY22 CARS project. 4. Enhance the CARS/warning sign map and PowerBI spreadsheet tool according to KYTC feedback. 5. Research the use of AI and machine learning by other states to inventory roadway assets, including signage. Complete a pilot study on one county using machine learning and KYTC’s Photolog to inventory curve warning signs. 6. Estimate the level of effort and cost to develop a machine learning method using Photolog for keeping the existing horizontal alignment signing inventory “relatively current”. 7. Work with KYTC to develop a process for updates and data sharing.